Astley S J, Clarren S K
Department of Epidemiology, School of Public Health and Community Medicine, Seattle, Washington.
Alcohol Clin Exp Res. 1995 Dec;19(6):1565-71. doi: 10.1111/j.1530-0277.1995.tb01025.x.
The purpose of this study was to derive a multivariate, quantitative case definition of the fetal alcohol syndrome (FAS) facial phenotype from a dysmorphologist-derived gold standard and use it to develop an effective screening tool for identification of children at risk for FAS. The facial and physical features of a racially mixed group of children (0.2-10.0 years of age), evaluated by a single dysmorphologist in the University of Washington FAS Clinic, were used to determine which feature or set of features best differentiated between children with and without a diagnosis of FAS. The study population was divided into two groups balanced on gender, age at examination, race, diagnosis, and date of examination. Group 1 was used to identify the most differentiating feature(s), and group 2 was used to validate the differentiating capability of the feature(s). Group 1 included 97 children (20 with FAS and 77 without FAS). Group 2 included 97 children (19 with FAS and 78 without FAS). Discriminant analysis identified smooth philtrum, thin upper lip, and short palpebral fissures as the cluster of features that best differentiated children with and without FAS based on the discriminant function [D = 1.7953086 + 0.8116083 (thin upper lip) + 2.6411562 (smooth philtrum)-3.4073780 (% predicted right palpebral fissure length)]. Patients with a D-score > or = 1.5 were classified as at-risk for FAS (screen positive). Using this cut-off value for the D-score, children in group 1 were classified with 100% sensitivity (20 of 20 true positives) and 90.0% specificity (70 of 77 true negatives). The children in group 2 were classified with 100% sensitivity (19 of 19 true positives) and 87.3% specificity (68 of 78 true negatives). Across all 194 patients, sensitivity was 100% [95% confidence interval (97-100)] and specificity was 89% [95% confidence interval (85 to 93)]. Seventy-one percent (n = 12) of the 17 false-positives had a true classification of possible fetal alcohol effects. Sensitivity and specificity were unaffected by race, gender, and age through 10 years. The screening tool is effective at differentiating children with and without FAS as diagnosed by a single dysmorphologist (S.K.C) at the University of Washington FAS Clinic. Assessment of diagnostic interrater agreement between trained dysmorphologists and testing in other clinic populations will be needed to assess the tool's external validity.
本研究的目的是从畸形学家制定的金标准中得出胎儿酒精综合征（FAS）面部表型的多变量定量病例定义，并利用该定义开发一种有效的筛查工具，以识别有FAS风险的儿童。由华盛顿大学FAS诊所的一名畸形学家对一组种族混合的儿童（0.2至10.0岁）的面部和身体特征进行评估，以确定哪些特征或特征组合能最好地区分已诊断和未诊断为FAS的儿童。研究人群分为两组，在性别、检查时的年龄、种族、诊断和检查日期方面保持平衡。第1组用于识别最具区分性的特征，第2组用于验证这些特征的区分能力。第1组包括97名儿童（20名患有FAS，77名未患FAS）。第2组包括97名儿童（19名患有FAS，78名未患FAS）。判别分析确定平滑人中、薄上唇和短睑裂是基于判别函数[D = 1.7953086 + 0.8116083（薄上唇）+ 2.6411562（平滑人中）- 3.4073780（预测右睑裂长度百分比）]能最好地区分患有和未患FAS儿童的一组特征。D评分≥1.5的患者被分类为有FAS风险（筛查阳性）。使用这个D评分的临界值，第1组中的儿童分类敏感性为100%（20例真阳性中的20例），特异性为90.0%（77例真阴性中的70例）。第2组中的儿童分类敏感性为100%（19例真阳性中的19例），特异性为87.3%（78例真阴性中的68例）。在所有194名患者中，敏感性为100%[95%置信区间（97 - 100）]，特异性为89%[95%置信区间（85至93）]。17例假阳性中有71%（n = 12）的真实分类为可能的胎儿酒精影响。敏感性和特异性在10岁之前不受种族、性别和年龄的影响。该筛查工具能够有效地区分华盛顿大学FAS诊所由一名畸形学家（S.K.C）诊断为患有和未患FAS的儿童。需要评估训练有素的畸形学家之间的诊断评分者间一致性，并在其他临床人群中进行测试，以评估该工具的外部有效性。
